Question 18

Explanation

✅ Correct Answer:
By upgrading existing FinOps metadata to include deep Kubernetes hierarchy and AI-specific metadata like model version and workload type.

Explanation:

Workday’s implementation of a centralized AI platform addressed a key FinOps challenge: the fragmentation of AI-related costs across diverse teams, services, and infrastructure layers. To gain detailed cost visibility and support unit economics (e.g., cost per prediction or per AI model version), Workday enriched their FinOps data by incorporating AI-specific metadata.

This meant tagging workloads not just with basic resource data, but with granular attributes such as:

• Model version

• Workload type (e.g., training vs. inference)

• Kubernetes namespace, pod, and container details

This upgrade allowed Workday to track cost per model, attribute usage to specific teams, and optimize resources based on actual consumption patterns, which is crucial in AI workloads where usage can be highly dynamic.

Other approaches like focusing only on aggregate spend, separating AI from other infrastructure, or locking into a single cloud provider wouldn’t provide the nuanced visibility or flexibility needed. Workday’s success came from improving metadata quality and granularity, not restricting architecture or simplifying reporting structures. This approach aligns closely with the evolving FinOps best practices for managing modern AI workloads.